Publication detail

Performance Prediction Model of Bus-Based Shared Memory Architectures

DVOŘÁK, V. STAROBA, J.

Czech title

Model architektury se sdílenou pamětí a sběrnicí orientovaný na predikci výkonnosti

English title

Performance Prediction Model of Bus-Based Shared Memory Architectures

Type

article in a collection out of WoS and Scopus

Language

en

Original abstract

It is shown that cache-coherent bus-based multiprocessor simulation can be implemented using message passing and few shared variables, at least in the case of an atomic bus and known coherence protocols. Data request and write-back bus transactions are generating messages to a shared memory server process, shared variables are used only for synchronization. A change in their values by one process is visible simultaneously to other processes, where it triggers invalidation/update actions. Models of various locks and barriers are described and the simulation-based performance prediction using Transim tool is illustrated on the example of parallel FFT benchmark in OpenMP. Multiprocessor hw, sw, and mapping to one another is described in Transim language that supports synchronous message passing as well as shared variables. Accuracy of prediction (8 %) has been satisfactory in the benchmark under test and may continue to be so in other benchmarks.

Czech abstract

Je ukázáno, že multiprocesor se sběrnicí a koherentními pamětmi cache se dá simulovat pomocí zasílání zpráv a několika sdílených proměnných, alespoň v případě atomické sběrnice a známých protokolů koherence. Sběrnicové transakce žádostí o data a o zpětný zápis generují zprávy do obslužného procesu sdílené paměti, zatímco sdílené proměnné jsou použity pouze pro synchronizaci. Změna jejich hodnot jedním procesem je současně viditelná ostatním procesům, v nichž spouští akce zneplatnění nebo aktualizace. Jsou popsány modely rozmanitých zámků a bariér a predikce výkonnosti založená na simulaci s pomocí nástroje Transim je ilustrována na příkladu paralelní testovací úlohy FFT v OpenMP. Hardware, software a mapování sw na hw je popsáno v jazyce Transim, který podporuje synchronní zasílání zpráv stejně jako sdílené proměnné. Přesnost predikce  (8 %) je pro testovanou úlohu uspokojivá a to může platit i pro další úlohy.

English abstract

It is shown that cache-coherent bus-based multiprocessor simulation can be implemented using message passing and few shared variables, at least in the case of an atomic bus and known coherence protocols. Data request and write-back bus transactions are generating messages to a shared memory server process, shared variables are used only for synchronization. A change in their values by one process is visible simultaneously to other processes, where it triggers invalidation/update actions. Models of various locks and barriers are described and the simulation-based performance prediction using Transim tool is illustrated on the example of parallel FFT benchmark in OpenMP. Multiprocessor hw, sw, and mapping to one another is described in Transim language that supports synchronous message passing as well as shared variables. Accuracy of prediction (8 %) has been satisfactory in the benchmark under test and may continue to be so in other benchmarks.

Keywords in English

Bus-based multiprocessor, message passing model, performance prediction, multiprocessor simulation, OpenMP

RIV year

2002

Released

22.04.2002

Publisher

Marq software s.r.o.

Location

Ostrava

ISBN

80-85988-71-2

Book

Proceedings of 36th International Conference MOSIS'02 Modelling and Simulation of Systems

Pages from–to

273–280

Pages count

8

BIBTEX


@inproceedings{BUT10007,
  author="Václav {Dvořák} and Jiří {Staroba},
  title="Performance Prediction Model of Bus-Based Shared Memory Architectures",
  booktitle="Proceedings of 36th International Conference MOSIS'02 Modelling and Simulation of Systems",
  year="2002",
  month="April",
  pages="273--280",
  publisher="Marq software s.r.o.",
  address="Ostrava",
  isbn="80-85988-71-2"
}